Claude (MCP)

The Telescope extension for Claude Desktop lets you ask questions about your portfolio in plain language, such as "Which of our properties in Bergen have a relevant flooding risk?" Claude looks up the answer in Telescope with your API key and explains it, builds tables, or drafts text from it.

What is MCP?

The Model Context Protocol (MCP) is an open standard for connecting AI assistants to other systems. The Telescope extension is an MCP server packaged as an .mcpb file (an MCP Bundle), which Claude Desktop installs in a few clicks. Nothing else needs to be installed.

How it works

┌────────────────┐   tools    ┌─────────────────────┐  X-API-Key  ┌───────────────┐
│ Claude Desktop │ ─────────▶ │ Telescope extension │ ──────────▶ │ Telescope API │
│ (you ask)      │ ◀───────── │ (on your computer)  │ ◀────────── │ /external/    │
└────────────────┘   results  └─────────────────────┘    JSON     └───────────────┘

1. When Claude Desktop starts the extension, it asks Telescope which operations your API key may use, and loads their descriptions from the API's OpenAPI schema.
2. Each allowed operation becomes a tool that Claude can call.
3. When you ask a question, Claude decides which tools to call and with which filters. The extension makes the request to Telescope with your key and hands the JSON back to Claude.
4. Claude answers from that data.

Because the tools come from Telescope when the extension starts, new tools appear after a restart of the extension, without installing a new file.

Available tools

Tool	What Claude can do with it	Kind
portfolios-list	List your root portfolios	Read
portfolios-retrieve	Open a portfolio with its child portfolios and properties	Read
properties-list	List and search properties, filtered by hazard, exposure, relevance and biodiversity	Read
properties-retrieve	Get one property with all its hazard and biodiversity results, buildings and energy certificates	Read
properties-create	Add a Norwegian property by matrikkelnummer, or update the one with the same reference ID	Write

properties-create only appears when creating is enabled for your organization. Read tools are marked read-only, so Claude Desktop can treat them differently from the write tool.

The tools work exactly like the API: same data, same filters, same organization scope. See Core concepts for what the results mean.

Permissions and data

• Claude sees what your key sees. That is your organization's data in Telescope, nothing more.
• You approve tool use. Claude Desktop asks before it calls a tool, unless you have chosen to always allow that tool. Consider keeping approval on for properties-create.
• Results become part of the conversation. The data a tool returns is sent to Claude to answer your question. Use the extension in line with your company's policy for AI tools.
• Your key stays on your computer. Claude Desktop stores it as a sensitive setting, and the extension only sends it to Telescope.
• No deleting. The extension can't delete anything. The write tool can update an existing property when its reference ID matches, so read what Claude plans to send before approving it.

Next steps

• Install the extension
• See what to ask